Release Notes

🌟 2025.07.29

🔹 Latest Support: glm-4.5, glm-4.5-air, glm-4.5-x, glm-4.5-airx, glm-4.5-flash

glm-4.5: Flagship model with 355B total parameters and 32B active parameters, designed for agentic applications, supporting hybrid reasoning modes and excelling in complex reasoning, tool calling, and web browsing.

glm-4.5-air: Cost-effective model with 106B total parameters and 12B active parameters, maintaining strong performance while significantly reducing costs, ideal for resource-sensitive applications.

glm-4.5-x: High-performance model optimized for ultra-fast inference and powerful reasoning capabilities, delivering millisecond-level response times for scenarios requiring speed and logic.

glm-4.5-airx: Lightweight yet powerful model combining Air's cost advantages with X's speed benefits, offering the perfect balance between performance and efficiency.

glm-4.5-flash: Efficient multi-purpose model with high generation speed, specifically optimized for coding and reasoning tasks, suitable for developers getting started and rapid prototyping.
-Follows the OpenAI chat standard format, see details: CometAPI Chat Documentation

🌟 2025.07.25

🔹 New Model: gemini-2.5-pro-all, gemini-2.5-flash-all, gemini-2.5-pro-deepsearch, gemini-2.5-flash-deepsearch, deepseek-r1t2-chimera

gemini-2.5-pro-all: A multimodal version of the Gemini model,

supporting analysis of files, videos, and images, as well as image generation and real-time web access.

gemini-2.5-flash-all: A multimodal version of the Gemini model,

supporting analysis of files, videos, and images, as well as image generation and real-time web access.

gemini-2.5-pro-deepsearch: A deep search model with enhanced deep search and information retrieval capabilities

ideal for complex knowledge integration and analysis.

gemini-2.5-flash-deepsearch: A deep search model combining the rapid performance of the Flash model with advanced deep search capabilities for fast, in-depth information discovery.

deepseek-r1t2-chimera: A 671B parameter Mixture-of-Experts (MoE) text generation model merged from DeepSeek-AI's R1-0528, R1, and V3-0324, supporting a context of up to 60k tokens.
-Follows the OpenAI chat standard format, see details: CometAPI Chat Documentation

🌟 2025.07.24

🔹 qwen3-coder-plus

qwen3-coder-plus: Focused on code generation, understanding, and optimization, excels in complex programming tasks.

🔹 qwen3-coder-plus-2025-07-22

qwen3-coder-plus-2025-07-22: Optimized version from 2025-07-22, stable and reliable, suitable for production.

🔹 qwen3-coder-480b-a35b-instruct

qwen3-coder-480b-a35b-instruct: Flagship model with 480 billion parameters, MoE architecture, capable of handling extremely complex programming.

Those models follows the openai chat standard format call, specific reference:
https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.07.18

🔹 Suno v4.5+

Suno v4.5+: v4.5+ has richer sounds, new creation methods, and a maximum length of 8 minutes. This website currently supports Suno 4.5+. Please change the request parameter mv to chirp-bluejay.

The above model follows the suno format, please refer to: https://apidoc.cometapi.com/api-13851480

🌟 2025.07.17

CometAPI supports Midjourney uploading masked images for local modifications

Refer to: https://apidoc.cometapi.com/api-18989894

🌟 2025.07.16

🔹 kimi-k2-0711-preview

kimi-k2-0711-preview: Kimi K2 is a large-scale mixed-expertise (MoE) language model developed by Moonshot AI.

with 1 trillion total parameters and 32 billion active parameters per forward pass. It is optimized for agent capabilities including advanced tool usage, inference, and code synthesis.

Kimi K2 performs well in a variety of benchmarks, especially in coding (LiveCodeBench, SWE-bench), reasoning (ZebraLogic, GPQA) and tool usage (Tau2, AceBench) tasks.

It supports long contextual inference with up to 128K tokens and features a novel training stack design that includes the MuonClip optimizer for stable large-scale MoE training.

The model follows the openai chat standard format call, specific reference:
https://apidoc.cometapi.com/chat-api-13851472

🌟 Since Google officially took down the gemini-2.5 model with version number on 7-15, the preview model has been forwarded to the official version; the gemini-2.5-pro model can be called as a priority. Click here to see the official notice: https://cloud.google.com/vertex-ai/generative-ai/docs/models/gemini/2-5-flash

🌟 2025.07.14

CometAPI now supports direct calls to the OpenAI API to process PDFs without uploading files by providing the URL of the PDF file.

For details on how to call it, see: https://apidoc.cometapi.com/api-18535147

🌟 As OpenAI has officially taken down the gpt-4.5 series models on 7-14; gpt-4.1 models can be called in priority Click here to see the official notice: https://platform.openai.com/docs/deprecations

🌟 2025.07.11

🚀 CometAPI supports Claude code!

• Add power to your development workflow. We're excited to announce that CometAPI now fully supports the powerful Claude Code.

• What does this mean for you?

• Top Artificial Intelligence features: Easily generate, debug and optimize code using models built specifically for developers.

• ⚙️ Flexible Model Selection: Our comprehensive range of models allows you to develop more seamlessly.

• Seamless Integration: APIs are always available. Integrate Claude Code directly into your existing workflow in minutes.

• Ready to build faster? Please click on the link below to make a call.

• click ：https://apidoc.cometapi.com/doc-1266358

🌟 2025.07.10

🔹 grok-4
🔹 grok-4-0709

grok-4，grok-4-0709: Currently supports text modal, with visual, image generation and other features coming soon. Extremely powerful technical parameters and ecological capabilities: Context Window: Supports up to 256,000 tokens of contextualization, ahead of mainstream models.

The model follows the openai chat standard format call, specific reference:
https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.07.04

Suno now supports stem separation, creating Persona, generating MP4 MV videos, getting WAV format files, and Timing: lyrics & audio timeline

Now supports Suno's full-track and single-track stem separation features, which can split your songs into up to 12 clean tracks—including vocals, drums, bass, etc., convenient for preview and download.

Create new Persona to generate Persona: singer style; generate music in different formats;

Note: Full-track stem separation feature is priced at 5 times the music generation cost; single-track billing is still being optimized, currently maintaining the same 5x pricing as full-track, will be charged at 1x base price in the future

For specific model usage, please refer to the above: https://apidoc.cometapi.com/api-18657316

🔹 veo3
🔹 veo3-pro
🔹 veo3-fast
🔹 veo3-frames
🔹 veo3-fast-frames
🔹 veo3-pro-frames

veo3,veo3-pro,veo3-fast: is the official Google's latest video generation model, the generated video with sound, the world's only video model with sound. veo3-frames,veo3-fast-frames,veo3-pro-frames Support first frame mode.

This model follows the OpenAI chat standard format for calls, refer to: https://apidoc.cometapi.com/api-18582532

🌟 2025.07.01

🔹 mj_fast_video

Midjourney video generation is now supported

Synchronized support for official website low-dynamic, high-dynamic, auto-generated, manual generation.

please click : https://apidoc.cometapi.com/api-18581293

🌟 2025.07.01

🔹 kling_image_expand

Now supports the Keyline API to expand the image.

please click : https://apidoc.cometapi.com/api-18584170

🌟 2025.06.25

🔹 black-forest-labs/flux-kontext-pro
🔹 black-forest-labs/flux-kontext-max
🔹 flux-kontext-pro
🔹 flux-kontext-max

black-forest-labs/flux-kontext-pro, black-forest-labs/flux-kontext-max:
The above two models follow the replicate call format; see details: https://apidoc.cometapi.com/api-16455857

flux-kontext-pro, flux-kontext-max:
The above two models follow the OpenAI chat standard call format; see details: https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.06.19

🔹 gemini-2.5-flash-lite-preview-06-17

gemini-2.5-flash-lite-preview-06-17: Large scale processing，Lower cost.

This model follows the OpenAI chat standard format for API calls, please refer to: https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.06.11

🔹 o3-pro
🔹 o3-pro-2025-06-10

o3-pro,o3-pro-2025-06-10: Supports web search, file analysis, visual input reasoning, Python programming, and personalized responses.
Compared to previous models, o3-pro shows significant improvements in clarity, completeness, instruction following, and accuracy.

This model adheres to the OpenAI v1/responses standard call format. For specific reference:

curl --location --request POST 'https://api.cometapi.com/v1/responses' \
--header 'Authorization: Bearer sk-xxxxxx' \
--header 'User-Agent: Apifox/1.0.0 (https://apifox.com)' \
--header 'Content-Type: application/json' \
--header 'Accept: */*' \
--header 'Host: api.cometapi.com' \
--header 'Connection: keep-alive' \
--data-raw '{
     "model": "o3-pro",
        "input": [{"role": "user", "content": "What’s the difference between inductive and deductive reasoning?"}]
    }'

🌟 2025.06.06

🔹 gemini-2.5-pro-preview-06-05

gemini-2.5-pro-preview-06-05: With native multimodal processing capabilities and a very long context window of up to 1 million words (Token), it provides unprecedented power for processing complex, long sequence tasks.

This model follows the OpenAI chat standard format for API calls, please refer to: https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.05.30

🌟 Gemini Models 1.5 series model off the shelf notice:

Due to the fact that Gemini no longer offers the 1.5 series officially, it will be phased out today;

Updated gemini-2.5-flash-preview-05-20 \ gemini-2.5-flash-preview-04-17 \ gemini-2.5-pro-preview-05-06 \ gemini-2.5-pro-preview-03-25 \ gemini-2.5-pro -preview-03-25 \ gemini-2.5-pro-exp-03-25 ;* gemini-2.5-pro-exp-03-25

Please continue to call the 2.5 series, the above models follow the openai chat standard format call, refer to: https://apidoc.cometapi.com/api-276386060

🌟 Notes on using gpt to generate images:

Since the interface gpt-4o-image is an interface realized by technical means, not an asynchronous interface, technically it can't achieve complete stability, so there will be instability.

If you have high stability requirements, we recommend using gpt-image-1, the official API call is more stable. Refer to the official images/generations format for the calling method, and https://api.cometapi.com/v1/images/generations for the url details.

Meanwhile, gpt-4o-image and gpt-image-1 support chat format, which can be called by technical means, please refer to the following url for details:
https://api.cometapi.com/v1/chat/completions

🌟 2025.05.29

🔹 deepseek-r1-0528

deepseek-r1-0528: Advanced reasoning capabilities, large parameter scale, powerful performance, suitable for complex tasks.

This model follows the OpenAI chat standard format for API calls, please refer to: https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.05.23

🔹 claude-sonnet-4-20250514

claude-sonnet-4-20250514: An important model in the Claude 4 series developed by Anthropic, significantly improving coding and reasoning capabilities compared to its predecessor Claude Sonnet 3.7.

It can respond more precisely to user instructions and efficiently handle complex tasks. This model is suitable for applications requiring high performance and cost-effectiveness. We've also added cometapi-sonnet-4-20250514 specifically for use in Cursor.

🔹 claude-sonnet-4-20250514-thinking

claude-sonnet-4-20250514-thinking: An important model in the Claude 4 series developed by Anthropic, significantly improving coding and reasoning capabilities compared to its predecessor Claude Sonnet 3.7.

🔹 claude-opus-4-20250514

claude-opus-4-20250514: Opus 4 is Anthropic's most advanced model, acclaimed as the world's best coding model.

It excels in handling complex, long-running tasks and intelligent agent workflows, particularly suitable for applications requiring high autonomy and intelligence.

🔹 claude-opus-4-20250514-thinking

claude-opus-4-20250514-thinking: Opus 4 is Anthropic's most advanced model, acclaimed as the world's best coding model.

It excels in handling complex, long-running tasks and intelligent agent workflows, particularly suitable for applications requiring high autonomy and intelligence.

This model follows the OpenAI chat standard format for calls, refer to: https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.05.07

🔹 Suno v4.5

Suno v4.5: v4.5 has more expressive music and richer vocals, designed to enhance the user's expression and intuition in music creation. This site now supports Suno 4.5, change the request parameter mv to chirp-auk

The above model follows the suno format, please refer to: https://apidoc.cometapi.com/api-13851480

🌟 2025.04.29

🔹 qwen3-235b-a22b

qwen3-235b-a22b: This is the flagship model of the Qwen3 series, with 235 billion parameters, utilizing a Mixture of Experts (MoE) architecture.

Particularly suitable for complex tasks requiring high-performance inference, such as coding, mathematics, and multimodal applications.

🔹 qwen3-30b-a3b

qwen3-30b-a3b: With 30 billion parameters, it balances performance and resource requirements, suitable for enterprise-level applications.

This model may use MoE or other optimized architectures, applicable for scenarios requiring efficient processing of complex tasks, such as intelligent customer service and content generation.

🔹 qwen3-8b

qwen3-8b: A lightweight model with 800 million parameters, designed specifically for resource-constrained environments (such as mobile devices or low-configuration servers).

Its efficiency and fast response capability make it suitable for simple queries, real-time interaction, and lightweight applications.

These models follow the OpenAI Chat standard format for calls. For specific details, please refer to:
https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.04.27

🔹 gpt-image-1

gpt-image-1 introduces native multimodal models to the API, built on GPT-4o's image generation capabilities, designed to provide developers with a powerful and flexible tool for generating high-quality, diverse images.

Features: High-fidelity images; diverse visual styles; rich world knowledge; consistent text rendering; unlocking practical applications across multiple domains.

This model follows the openai v1/images/generations format for calls, see details at: https://apidoc.cometapi.com/images-api-13851474 ;Here's an example of input parameters:

{
    "model": "gpt-image-1",
    "prompt": "A cute baby sea otter",
    "n": 1,
    "size": "1024x1024"
}

🌟 2025.04.20

🔹 gemini-2.5-flash-preview-04-17

gemini-2.5-flash-preview-04-17, Gemini 2.5 Flash is an AI model developed by Google, designed to provide developers with fast and cost-effective solutions, especially suitable for applications requiring enhanced reasoning capabilities.

According to the Gemini 2.5 Flash preview announcement, the model's preview version was released on April 17, 2025, supports multimodal input, and has a context window of up to 1 million tokens.

This model follows the OpenAI chat standard format for calling, refer to:https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.04.17

🔹 o4-mini
🔹 o4-mini-2025-04-16

o4-mini, o4-mini-2025-04-16: A smaller, faster, and more economical model, research shows it performs well in mathematics, coding, and visual tasks, designed to be efficient and responsive, suitable for developers. Released on April 16, 2025.

🔹 o3
🔹 o3-2025-04-16

o3, o3-2025-04-16: A reflective generative pre-trained transformer (GPT) model designed to handle problems requiring step-by-step logical reasoning.

Research shows it excels at mathematics, coding, and scientific tasks. It can also use tools such as web browsing and image generation, with a release date of April 16, 2025.

The above models follow the OpenAI chat standard format for calls, refer to: https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.04.15

🔹 gpt-4.1

gpt-4.1: Major advancements in coding and instruction following; GPT-4.1 has become the leading model for coding.

Long context: On Video-MME, a benchmark for multimodal long context understanding, GPT-4.1 has created a new state-of-the-art result.

The GPT-4.1 model series delivers superior performance at lower cost.

🔹 gpt-4.1-mini

gpt-4.1-mini: Represents a significant leap in small model performance, even outperforming GPT-4o on many benchmarks.

It matches or exceeds GPT-4o in intelligence assessment while reducing latency by nearly half and costs by 83%.

🔹 gpt-4.1-nano

gpt-4.1-nano: Features a larger context window—supporting up to 1 million context tokens

And can better utilize this context through improved long context understanding. Has an updated knowledge cutoff date of June 2024.

These models follows the standard OpenAI chat format for API calls, for reference see: https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.04.14

🔹 grok-3-deepersearch

grok-3-deepersearch: Features high data timeliness, excellent interactive experience, and thorough search thinking process; comprehensive webpage aggregation.

This model follows the OpenAI chat standard format for API calls, refer to: https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.04.13

🔹 gemini-2.0-flash-exp-image-generation

This model supports conversation while enabling image generation and editing capabilities, outputting high-definition images.

This model follows the OpenAI chat standard format for API calls, refer to: https://apidoc.cometapi.com/api-15928299

🌟 2025.04.10

🔹 grok-3-fast
🔹 grok-3-fast-latest

grok-3-fast, grok-3-fast-latest: grok-3 and grok-3-fast use exactly the same underlying model and provide the same response quality. However, grok-3-fast is served on faster infrastructure, delivering response times that are much quicker than the standard grok-3.

This model follows the OpenAI chat standard format for API calls, refer to: https://apidoc.cometapi.com/chat-api-13851472

🔹 grok-3-mini
🔹 grok-3-mini-latest

grok-3-mini, grok-3-mini-latest: A lightweight model that thinks before responding. Fast, intelligent, and ideal for logic-based tasks that don't require deep domain knowledge. The original thought traces are accessible.

This model follows the OpenAI chat standard format for API calls, refer to: https://apidoc.cometapi.com/chat-api-13851472

🔹 grok-3-mini-fast
🔹 grok-3-mini-fast-latest

grok-3-mini-fast, grok-3-mini-fast-latest: grok-3-mini and grok-3-mini-fast use exactly the same underlying model and provide the same response quality. However, grok-3-mini-fast is served on faster infrastructure, delivering response times that are much quicker than the standard grok-3-mini.

This model follows the OpenAI chat standard format for API calls, refer to: https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.04.07

🔹 llama-4-maverick

llama-4-maverick, a high-capacity multimodal language model from Meta, supports multilingual text and image inputs and generates multilingual text and code output in 12 supported languages.

Maverick is optimized for visual language tasks and has instructions tuned for assistant-like behavior, image reasoning, and generic multimodal interaction.

Maverick features native multimodal early fusion and 1 million labeled context windows.

Maverick is released on April 5, 2025 under the Llama 4 Community License for research and commercial applications requiring advanced multimodal understanding and high model throughput.

The model follows the openai chat standard format call,cf:https://apidoc.cometapi.com/chat-api-13851472

🔹 llama-4-scout

llama-4-scout, is a mixed-expertise (MoE) language model developed by Meta. It supports native multimodal input (text and images) and multilingual output (text and code) for 12 supported languages.

Designed for assisted interaction and visual reasoning, Scout uses 16 experts per forward pass, a context length of 10 million words, and a training corpus of about 40 trillion words.

Designed for high efficiency and local or commercial deployment, llama-4-scout employs early fusion technology for seamless modal integration.

It is command-tuned for multilingual chat, subtitling, and image comprehension tasks.

It is released under the Llama 4 Community License, with last training data as of August 2024 and a public release on April 5, 2025.

The model follows the openai chat standard format for calls, cf:https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.03.29

🔹 gpt-4o-all

gpt-4o-all has support for ChatGPT's latest generated image mode

The model follows the openai chat standard format for calls, cf:https://apidoc.cometapi.com/api-15928299

🔹 gpt-4o-image

gpt-4o-image This model is dedicated to image generation and editing, which enables image style conversion, preservation of original image features, superb consistency, and output of high-definition images.

The model follows the openai chat standard format for calls, cf:https://apidoc.cometapi.com/api-15928299

🌟 2025.03.27

🔹 gemini-2.5-pro-exp-03-25

Features native multimodal processing capabilities, with an extensive context window of up to 1 million tokens, providing unprecedented powerful support for complex, long-sequence tasks.

The model follows the openai chat standard format for calls, cf:https://apidoc.cometapi.com/chat-api-13851472

🔹 gemini-2.5-pro-preview-03-25

According to Google's data, Gemini 2.5 Pro demonstrates particularly outstanding performance in handling complex tasks.

The model follows the openai chat standard format for calls, cf:https://apidoc.cometapi.com/chat-api-13851472

🌟 2025.03.24

🔹 gpt-4.5-preview-2025-02-27

Preview Version: Showcasing the latest features of GPT-4.5, providing enhanced understanding and generation capabilities, suitable for various tasks, improving user experience.

🔹 gpt-4.5-preview

Preview Version: Deeply optimized algorithms and performance, delivering ultra-fast responses and precise outputs, perfectly suited for efficient decision-making scenarios.

🔹 gpt-4.5

Professional Standard Version: Stable and reliable, combining rich expression and multi-task processing capabilities, suitable for wide applications including business, education, creative, and technical fields.

🌟 2025.02.20

🔹 claude-3-7-sonnet-thinking

Advanced model designed for complex reasoning and creative thinking, unleashing unlimited possibilities, empowering breakthrough problem-solving and innovation.

🔹 claude-3-7-sonnet-20250219

High-end version integrating the latest technological breakthroughs, handling complex tasks with superior performance, providing intelligent innovative solutions for users.

🔹 cometapi-3-7-sonnet

Outstanding multi-domain processing expert, delivering precise and smooth output experience, easily tackling various professional challenges.

🔹 cometapi-3-7-sonnet-thinking

Equipped with revolutionary algorithm architecture, significantly enhancing deep analysis and complex task management capabilities, making thinking more thorough and comprehensive.

🔗 Usage Guide:

✅ All models have been added to the default group, allowing you to flexibly call them according to different usage scenarios and requirements, easily integrate, and maximize their application value.

🛠 Quick Start:

Simple integration into your system unlocks powerful capabilities. Fully utilize each model's unique advantages to meet professional needs across different domains.

🔥 Experience the revolutionary performance improvements these breakthrough models bring right now! 🔥

For professional support or detailed consultation, please contact our customer service team or visit our technical documentation center. We look forward to your valuable feedback!

🌟 2025.07.29#

🌟 2025.07.25#

🌟 2025.07.24#

🌟 2025.07.18#

🌟 2025.07.17#

🌟 2025.07.16#

🌟 2025.07.14#

🌟 2025.07.11#

🌟 2025.07.10#

🌟 2025.07.04#

🌟 2025.07.01#

🌟 2025.07.01#

🌟 2025.06.25#

🌟 2025.06.19#

🌟 2025.06.11#

🌟 2025.06.06#

🌟 2025.05.30#

🌟 2025.05.29#

🌟 2025.05.23#

🌟 2025.05.07#

🌟 2025.04.29#

🌟 2025.04.27#

🌟 2025.04.20#

🌟 2025.04.17#

🌟 2025.04.15#

🌟 2025.04.14#

🌟 2025.04.13#

🌟 2025.04.10#

🌟 2025.04.07#

🌟 2025.03.29#

🌟 2025.03.27#

🌟 2025.03.24#

🌟 2025.02.20#

🔗 Usage Guide:#

🛠 Quick Start:#

🌟 2025.07.29

🌟 2025.07.25

🌟 2025.07.24

🌟 2025.07.18

🌟 2025.07.17

🌟 2025.07.16

🌟 2025.07.14

🌟 2025.07.11

🌟 2025.07.10

🌟 2025.07.04

🌟 2025.07.01

🌟 2025.07.01

🌟 2025.06.25

🌟 2025.06.19

🌟 2025.06.11

🌟 2025.06.06

🌟 2025.05.30

🌟 2025.05.29

🌟 2025.05.23

🌟 2025.05.07

🌟 2025.04.29

🌟 2025.04.27

🌟 2025.04.20

🌟 2025.04.17

🌟 2025.04.15

🌟 2025.04.14

🌟 2025.04.13

🌟 2025.04.10

🌟 2025.04.07

🌟 2025.03.29

🌟 2025.03.27

🌟 2025.03.24

🌟 2025.02.20

🔗 Usage Guide:

🛠 Quick Start: